Additional Visual Norm rule to replaces ZWNJ with ZWJ for Bengali YA-…

…phalaa only, as its usage with ZWNJ is common in the web. PiperOrigin-RevId: 576506902
google-research · Oct 25, 2023 · 4de73d4 · 4de73d4
1 parent a37ec88
commit 4de73d4
Show file tree

Hide file tree

Showing 2 changed files with 9 additions and 0 deletions.
diff --git a/nisaba/scripts/brahmic/data/Beng/visual_rewrite.textproto b/nisaba/scripts/brahmic/data/Beng/visual_rewrite.textproto
@@ -55,3 +55,9 @@ item {
   uname: ["AU LENGTH MARK", "SIGN E"] raw: "ৗে"
   to_uname: "SIGN AU" to_raw: "ৌ"
 }
+
+# Replaces ZWNJ with ZWJ for YA-phalaa.
+item {
+  uname: ["RA", "ZERO WIDTH NON-JOINER", "VIRAMA", "YA"] raw: "র‌্য"
+  to_uname: ["RA", "ZERO WIDTH JOINER", "VIRAMA", "YA"] to_raw: "র‍্য"
+}
diff --git a/nisaba/scripts/brahmic/testdata/visual_norm.textproto b/nisaba/scripts/brahmic/testdata/visual_norm.textproto
@@ -132,6 +132,9 @@ rewrite { rule: "BENG" input: "অ্যাড" output: "অ্যাড" }
 rewrite { rule: "BENG" input: "এ্যালজিও" output: "এ্যালজিও" }
 
 # YA-phalaa/sign
+# <RA, ZWNJ, VIRAMA, YA> → <RA, ZWJ, VIRAMA, YA>
+rewrite { rule: "BENG" input: "র‌্যাল" output: "র‍্যাল" }
+# ZWJ in <RA, ZWNJ, VIRAMA, YA> is preserved.
 rewrite { rule: "BENG" input: "র‍্যাল" output: "র‍্যাল" }
 rewrite { rule: "KNDA" input: "ರ‍್ಯಂಡ್" output: "ರ‍್ಯಂಡ್" }